ul 2004
e built
d of a
as well
eed fill
ncrease
ing built
. whose
it means
999 for
T
E
SITY
81-1009,
61-80%
41-60%
21-40%
0-20%
been set.
ver total
D = 40-
JNOS 10
14.44
18.39
26.35
27.19
13.63
ked in A
| Landsat
idsat and
5 is quite
vs higher
or classes
| Landsat
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
SPOT A SPOTE SPOT C SPOT D SPOT E
80 80 80 4 80 80 4
40 40 40 Is 40 | 40
0 EE 0 0 | 0 | 0
ABCDE ABCDE ABCDE
ABCDE ABCDE
IKONOS A IKONOS B
A:B CD E
IKONOS C IKONOSD IKONOS E
;| 904
45 4
ABCDE
ABCDE
Figure 8. Graphic representation of cross table 3
[ IKONOS. 5
In blue:
Distribution of each
density class of Spot
into IKONOS. 5
classes.
(percentage values)
Distribution of each
density class of
IKONOS, 5 into Spot
classes
(percentage values)
|o 0 0.75
TOT (9.82 |18.11 [22.97
100 (100 | 100
Table 4. Cross table
SPOT A SPOT E SPOT C SPOT D SPOT E
90 90 en 55 90
alis File 45 +| 45 ani
al og eld] 9 T 0 4 |
ABCDE ABCDE ABCDE ABCDE ABCDE
| IKONOS
[^ —R G D | TOT
SPOT | | | |
A | &75 411] al al ol 10.6] In blue :
| ö 9 0 Distribution of each
pU TT FY A Lu 18 density class of Spot into
seu) ane) REI Mie] eia ues. , 3
| ss am © IKONOS classes.
; | 6.67! 10.45} 68.08 14.8] 0] 100 (percentage values)
C | Or 1,59 6.25 15.64 0j 21.42
| 0| 0X 307 4XOB — 0
| Ol 7.41 29.09| £3.51 0| 100
5 | al nai SES] FI] 151 ES ^ WIRE
i 0i 438 1 e 4 zd In ink; . =
| M | 100 Distribution of each
E | jj 21.23 density class of
| . onl IKONOS into Spot
[TOT | 7.8] 791] 20.51] 27.79] 35.89] 100 classes
| mE fof m too 106 (percentage values)
IKONOS 5 C IKONOS 5D IKONOS 5E
6 6 ‚| 96
E t |.
HN Leo]
ABCDE ABCDE
ABCDE ABCDE
Figure 9. Graphic representation of cross table 4
To provide insight about changes in scale, we analyse - by
means of two by two overlays of five maps of density - the
questions such as: to which density class of IKONOS go
pixels of the class A of Spot and so on (tables 3 and 4, figures
8 and 9).
From this analysis it emerges that:
1. most (percentage more than 50%) of the pixels of a given
class of IKONOS pass to the higher density class in Landsat
and Spot.
2. Landsat and Spot density classes distributions are similar
even spatially.
3. IKONOS_5 and Spot density classes distributions are
similar: nearly all pixels belong to the same classes of density.
distribution of density classes of a given image in the classes of
another image at a different resolution. This allows us to answer
469
and this mainly for classes A (92%) and B (63%).
The same remarks are valid for IKONOS_10 with regard to
Landsat.
6. THIRD WORK
In the study area analysed above, according to urban studies,
five different sub-areas displayed and drawn in the fig. 14 can
be identified: 1_compact core; 2_suburban area; 3_
connection area; 4_scattered settlements; 5_ industrial area.
The issue under consideration is to what extent landscape
indexes can be used to characterize and identify different
urban structures. Such analysis is set forth on the five images
mentioned above, on which the same five sub-areas have been
drawn.
fel ia ’ 2
A OM a AT. y SE
Figure 10. Study area and 5 sub-areas drown on map from
IKONOS
; S
DE E M. zl
6.1 Statistics of landscape
Results from spatial statistics mentioned in paragraph 3, bring
out some interesting points:
1. Areas 1 and 2 show low values of PD (5 and 8) and ED
(45 and 106) in Spot and Landsat, while they are high
(PD=447; ED=747) in IKONOS ( table 5, figures 11 and 12).
MA is high in Spot and Landsat, very low in IKONOS.
Statistics clearly demonstrate two different ways to represent
concentrated urban areas: uniform and compact in Landsat
and Spot; a dense set of single elements in IKONOS.
For these areas IKONOS. 5 and IKONOS 10 provide a
representation quite similar to Landsat and Spot.
